2022 IEEE 2nd Mysore Sub Section International Conference (MysuruCon)

Anomaly Alert System using CCTV surveillance

Manu Y M Ravikumar G K Shashikala S V
Department of R&D Department of Computer Science and Department of Computer Science and
BGS Institute of Technology, Engineering Engineering
Adichunchanagiri University BGS Institute of Technology, BGS Institute of Technology,
B.G Nagara, Mandya Adichunchanagiri University Adichunchanagiri University
2022 IEEE 2nd Mysore Sub Section International Conference (MysuruCon) | 978-1-6654-9790-9/22/$31.00 ©2022 IEEE | DOI: 10.1109/MysuruCon55714.2022.9972363

[email protected] B.G Nagara, Mandya B.G Nagara, Mandya

[email protected], [email protected]

Abstract- Even with its probable risks and capacity to in the backdrop leading to a shortage of oversight. Even if
quickly spread, fire poses a huge danger to life. Fire detection we've been watching an area with CCTV, if the cop's
techniques, particularly vision-based mechanisms, identify superiors are caught unawares and do not act quickly when
flames before any damage or devastation occurs. We suggested the API is captured by CCTV and so cannot be refuted, the
a fire detection method image processing and analysis fire will continue to spread. Since CCTV could only oversee,
approaches in this research, which is appropriate with a mechanism or reasoning ability, image processing should
surveillance systems such as CCTV. We'll make a fire sensor be added to the CCTV functionality. As a result, image
out of a Raspberry Pi and a Raspberry Pi camera, then use processing refers to a variety of signal analysis in which the
Image Processing to acquire real-time recognition results and
user can interact, such as a photo or video, and the outcome
evaluate the accurate fire image. The fire image will then be
examined with Image Processing, and the outcome will be sent
is a picture or a set of image-related features or attributes.
to the user as a caution (alert) that there is a fire in the area. When image processing is combined with camera
surveillance, the image analysis will proceed to nurture the
Keywords—Fire detection, Vision-based system, Image picture that ignites and exhibit it on the panel surveillance.
Processing CCTV, Raspberry Pi

II. LITERATURE SURVEY

I. INTRODUCTION
Owing of safety and liability concerns, warehouses have a In this paper they have employed a computer vision-
surveillance system of cameras. A costly threat monitoring based system to detect fire and transmit an alarm to a distant
system, comprising of fire alarm systems and smoke sensing fire brigade. Because the area seems to be under continual
devices, is generally installed in factories. In the event of a monitoring, this alarm is provided not just as an urgency
fire, various technologies may not have been very functional. signal but rather as a video. We can determine how many
Furthermore, alerts are often owing to cigarette or incense individuals are still inside room using this footage, and the
smoke, among other things. As a result, we designed an fire department may dispatch adequate responders to the site
application that could be used with a current surveillance based on the video[1].We limit employing flames and smoke
system to analyse a live stream and identify fire. detectors, which might cause false warnings. If our
The proposal's goal is to address the current issue of faulty technology generates a fake alert, we may review the video
fire protection systems in commercial warehousing. The to validate it. In this research, we offer a fire approach to
objective of the study is to use security cameras to identify detect based on OpenCV and Raspberry Pi that detects fire
and manage the onset of fires. Because the cameras seem to and sends alerts as a siren within the same facility[2], as well
be in existence, this approach minimizes the risks of error as a brief video to a distant fire alarm control unit[3].
rates, proposed system a cost-effective and quick way of fire The author proposes a smarter method for recognizing
detection. Since we all understand, fires are quite prevalent the boundaries of sparks by eliminating all disturbance from
and manufacturing sites, and they can be created by human the sparks. According to certain study, current approaches do
or machine fault in operation, as well as a short-circuit or not prioritize the consistency and transparency of both the
shorted power. Spontaneous fires usually occur in areas spark and fire margins[4]. The suggested approach
where there is less control, like distribution centre. Initially, recognizes the flame/continuous fires and obvious
there was presumably only a little flicker of fire in the area boundaries. This method recognizes the contours of entities
but owing to the absence of vigilance and the speed with and the limits between them and the image's backdrop. The
which the fire was anticipated, the sparks dispersed across application's efficacy and resilience were demonstrated in
the room, causing the fire to spread. experiments using various flame photos[5].
With the digital revolution, technologies such as video The method taken in this study can detect fire as well as
surveillance, sometimes known as CCTV, become available. provide employees a warning. A Raspberry Pi 3 with
CCTV is being used to watch or supervise areas where there detectors and a camera was utilized. Gas leaks are detected
is very little monitoring, and even if there are any such by a gas sensor, entry is detected by a PIR sensor, and fire is
instances, the flare may be detected quickly before it spreads detected by a flame sensor[6]. The detectors are continually
and causes a fire. Even if there is no permanent CCTV sensing and transferring data to an internet web browser over
surveillance, it is unusual that the name fires. It will operate Wi-Fi. When a fire occurs because of an intrusion, the

978-1-6654-9790-9/22/$31.00 ©2022 IEEE

1
Authorized licensed use limited to: Shah and Anchor Kutchhi Engineering College. Downloaded on August 22,2024 at 08:01:29 UTC from IEEE Xplore. Restrictions apply.
 camera takes a snapshot, and the system sends an email Camera, as well as use the idea of Image Processing to
notice to the enterprise owner including an image of the obtain real-time recognition data and evaluate the accurate
afflicted area[7]. image of fire. The camera will keep monitoring the
The suggested system begins with picture pre-processing circumstances in the area if the gadget is functioning.
and segmentation, which incorporates algorithms for corner
detection and thresholding. The parameters are taken using
the Histogram of Oriented Gradient (HOG) technique and
the Gray Level Co-occurrence Matrix (GLCM) algorithm.
Support Vector Machines (SVMs) are a classification
technique[8]. After the image analysis portion, the
recognized fire intensity is fed into the microcontroller unit,
where the LCD shows the fire intensity as regular, moderate,
or extreme, as well as sending an information to the recipient
in charge. A buzzer is used to signal if the fire is light or
serious. As a result, the technology saves lives and detects
fires early in various regions[9].
This research proposes a method i.e. video-based flame
smoke detection technique in a real-time environment that
may be used with an autonomous surveillance system for
timely alarms[10]. The motion history feature extraction is
used in the subsequent processing phases of our real-time
approach to capture the probable location of fire smoke in a
clip and then assess the harmonic, spatial, and temporal
information of the fire smoke areas in the picture frames[11].
When using HSI color spaces, the harmonic probability
distribution is depicted by contrasting the fire smoke
histogram - based model. Analyzing the fire smoke
turbulence dynamics with the periphery and area relationship
determines the physical posterior distribution[12]. By
isolating the fluctuating zone with signalized intersection and
isolating the counterfeit artifacts out from fire smoke zone,
the periodic probability distribution may be displayed[13].
Then, at a phenomenal speed, the continuously adaptive
mean shift (CAMSHIFT) visual tracking method is used to
offer real-time feedback on the fire smoke location[14]. The
suggested approach effectively detects fire smoke, according
Fig..1. Architecture Overview.
to research observations in a range of situations. [15], Fire
normally originates serious hazards. So, to prevent calamity
that arise in industries, construction field, and wild forest Fig 1 Tells Overall overview of the system. It consists of 3
zones, here image based technique that detects fire has main functionalities
become an important issue[16]. 1. Image acquisition
Detection of fire technology using OpenCV and 2. Fire detection using haar-cascade
Raspberry Pi to identify fire and give an alert as an alarm in 3. Alert if fire detected
the same building and a short video clip is sent to the system
i.e. called remote fire alarm control unit[17].
Through this video, user came to know how many people
are present within the room and the fire station might send
sufficient rescuers based on the video to rescue people[18]. Figure.1.
The main advantage is that we avoid using flame and smoke
sensors that might give false alarms[19]

III. PROPOSED METHHOD

The Raspberry Pi serves as the device's intelligence, and a

CCTV camera is utilized to fire alarms. Image analysis is Figure 2: System Analysis.
employed in this application to automatically detect and Fig..2. F Associative System.
notify the user[20].
Image processing is a method of taking a picture as an Fig. 2 Shows associate systems degree lysis is method by
intake and transforming it into image as an outcome using that private studies a system specified a systems specified
various approaches. This chance will allow us to build a fire Systems are often analyzed, modeled and chosen. Analytic
alarms device utilizing a Raspberry Pi and a Raspberry Pi

Authorized licensed use limited to: Shah and Anchor Kutchhi Engineering College. Downloaded on August 22,2024 at 08:01:29 UTC from IEEE Xplore. Restrictions apply.
 thinking comes and initiated for 3 reasons issues, Fig 4 Shows Use Case Diagram, Since video patterns are
opportunities and directives. too many, here we have sampled them with a sampling rate
of 100 frames. Sampled video patterns are used to trial the
achievement of our algorithm. Here we used the below
performance criterion. 1st the algorithm is applied on
sampled video patterns. Then the rectified fire frames are
measured manually, and we have compared them to the
original frames. If the method raises the alarm flag for a
frame with no fire in it, then there is a possibility of error
occurrence. The correct detection rate is on the order of
98.89%, which we assume to be an acceptable level for fire
detection. The 1.11% error in detection rate is due to camera
indiscriminate, and there is sudden change in lighting
Fig..3. Database Design. conditions, and also some kind of materials producing
different fire colors while they are in burning [22].
Fig 3 Shows Fire extinguisher Database Design i.e.
mandatory in every building according to the norms of
government. Generally, people don’t have the knowledge
how to use the fire extinguisher and even at times of critical
situation people will become panic, and doesn’t react to the
situation. Bottom-Level refining operation adds tasks such Fig..5. Block diagram of Edge Detection system.
as image prequel to reduce noise, contrast heightening,
image sharpening, etc. [21] In the last-level process, images Edge detection technique is used to detect the color variance
must be both output and input. in an video image. Block Diagram of Edge Detection
System is as shown in Fig 4.

Fig..6. Block diagram of motion detection

Motion detection is used to detect any occurrences of

movement in a sample video of sequential two frames.
Block diagram of motion detection system is as shown in fig
6.
considering all OCP descriptors , ,………,
in a trained crowd videos, proposed HBT method is used
with feature vector representation as shown in below
Equation.

min ,

≜ 2
(1)
!
"# ! $

Feature vector coding based proposed HBT method required

to be limited in a closed set by following equation.

≜ %
∈ ' ()* +. -. ∀/ (2)
1, … … , , 123 12
4 15

Fig..4. Use Case Diagram.

Authorized licensed use limited to: Shah and Anchor Kutchhi Engineering College. Downloaded on August 22,2024 at 08:01:29 UTC from IEEE Xplore. Restrictions apply.
 IV. RESULTS AND DISCUSSION
Detection of fire using the proposed technology is
demonstrated using the below images.

We proposed a fire identification algorithm based on image

processing techniques. And this algorithm uses RGB color
framework to identify the color of the fire which is mainly
comprehended by the intensity of the component R which is
red color [23]. The broadening of fire is detected using
Sobel edge detection. Finally, a color based segmentation
technique was applied based on the results from the first
technique and second method to rectify the region of interest
(ROI) of the fire[24]. The algorithm works very well when
there is a fire outbreak. The overall accuracy of the work is
more than 90%, indicating the effectiveness and usefulness
of the algorithm. In future work, a real-time based algorithm
could be consider as it might increase the efficiency of the
algorithm which is currently 80.64%[25]. Fig 7, Fig 8 and
Fig 9 Shows the detected fire in rectangles.
Fig..9. Shows the fire detected in the rectangle box.

V. CONCLUSION

With the help of a camera and a Raspberry Pi, this method

can monitor fire in the property. It also distinguishes
between fire and the underlying operating system. As a
result, the mechanism accomplishes its goals. A lot of
cameras might be attached to a single Raspberry Pi in the
future to trigger an alarm in different areas of the same
structure.
Fig..7. Shows the fire detected in the rectangle box.
REFERENCES
[1] G. Sathyakala, V. Kirthika, B. Aishwarya, “Computer Vision Based
Fire Detection with a Video Alert System”, 2018 International
Conference on Communication and Signal Processing (ICCSP).
[2] Y. Kalpana, M. Padmaa, “An efficient edge detection algorithm for
flame and fire image processing”, 2014 International Conference on
Communication and Signal Processing.
[3] D. Deepa, P. Monica, V.S. Monisha, R. Mahalakshmi, “Design of an
IOT approach for Security Surveillance system for Industrial process
monitoring using Raspberry Pi”, 2019 IEEE International Conference
on Innovations in Communication, Computing and Instrumentation
(ICCI).
[4] D. Jayashree, S. Pavithra, G. Vaishali, J. Vidhya, “System to detect
fire under surveillanced area”, 2017 Third International Conference
on Science Technology Engineering & Management (ICONSTEM).
[5] Chao-Ching Ho, Tzu-Hsin Kuo, “Real-time video-based fire smoke
detection system”, 2009 IEEE/ASME International Conference on
Advanced Intelligent Mechatronics.
[6] Richard Szeliski, “Computer Vision: Algorithms and Applications”,
published in September 2010.
[7] Md Saifudaullah Bin Bahrudin, Rosni Abu Kassim, “Development of
fire alarm system using Raspberry Pi and Arduino Uno,” 2013
International Conference on Electrical, Electronics and System
Fig..8. Shows the fire detected in the rectangle box. Engineering, December 2013.
[8] Ahmed Imteaj, Tanveer Rahman, Muhammad Kamrul Hossain,
Mohammed Shamsul Alam, Saad Ahmad Rahat, “An IoT based fire
alarming and authentication system for workhouse using Raspberry Pi
3”, International Conference on Electrical, Computer and
Communication Engineering(ECCE), February 2017.
[9] Rajeshkumar Kallur, Prof. Rohini Kallur, “Fire detection with a
wireless video camera using Matlab”, International Journal of Recent
Trends in Engineering & Research”, volume 02, Issue 04, April 2016.

Authorized licensed use limited to: Shah and Anchor Kutchhi Engineering College. Downloaded on August 22,2024 at 08:01:29 UTC from IEEE Xplore. Restrictions apply.
 [10] Chandrakant Shrimantrao, Mahesh S K, Vivekanand M Bonal, “Fire [19] C. Szegedyet al., ‘‘Going deeper with convolutions,’’ in Proc. IEEE
detection system using Matlab”, International Journal for Research in Conf. Comput. Vis. Pattern Recognit., Jun. 2019, pp.1–9.
Applied Science & Engineering Technology, volume 5 Issue VII, July [20] A. Ullah, J. Ahmad, K. Muhammad, M. Sajjad, and S. W. Baik,
2017. ‘‘Action recognitioninvideosequencesusingdeepBi-
[11] Vijayalaxmi, B.Shravani, G.Sree Ram, “Fire detection and Alert directionalLSTMwithCNN features,’’ IEEE Access, vol. 6, pp. 1155–
system using image processing”, Global Journal of Advanced 1166,2018.
Engineering Technologies. Special Issue(CTCNSF-2014). [21] Chandrakant Shrimantrao, Mahesh S K, Vivekanand M Bonal, “Fire
[12] K.S.Shilpashree, Lokesha.H, Hadimani Shivakumar, “Implementation detectionsystem using Matlab”, International Journal for Research in
of Image Processing on Raspberry Pi”, International Journal of Applied Science &EngineeringTechnology, volume 5 IssueVII,
Advanced Research in Computer and Communication Engineering, July2017.
Vol 4, Issue 5, May 2015. [22] R.Dhanujalakshmi,B.Divya,C.Divya@sandhiya,A.Robertsingh,“Imag
[13] Ashwin Pajankar, “Raspberry Pi Computer Vision Programming”, eProcessingbasedfiredetectionsystemusingRaspberryPisystem”,SSRG
Published in May 2015. International Journal of Computer science and Engineering (JCSE),
[14] R.Dhanujalakshmi, B.Divya, C.Divya@sandhiya, A.Robertsingh, Volume 4Issued4 April 2017.
“Image Processing based fire detection system using Raspberry Pi [23] R.Dhanujalakshmi,B.Divya,C.Divya@sandhiya,A.Robertsingh,“Imag
system”, SSRG International Journal of Computer science and eProcessingbasedfiredetectionsystemusingRaspberryPisystem”,SSRG
Engineering (JCSE), Volume 4 Issued 4 April 2017. International Journal of Computer science and Engineering (JCSE),
[15] Prof. Amit Hatekar, Saurabh Manwani, Gaurav Patil, Akshat Parekh, Volume 4Issued4 April 2017.
“Fire detection on a surveillance system using Image processing”, [24] Prof. Amit Hatekar, Saurabh Manwani, Gaurav Patil, Akshat Parekh,
International Journal of Engineering Research & Technology(IJERT), “Fire detection on a surveillance system using Image processing”,
Vol.6 Issue 05, May-2020. International Journal of Engineering Research & Technology(IJERT),
[16] J. Yang, B.Jiang,B.Li,K.Tian,andZ.Lv, ‘‘Afastimageretrievalmethod Vol.6 Issue 05, May-2020.
designed for network big data,’’ IEEE Trans. Ind. Inform., vol. 13, no. [25] J. Yang, B.Jiang,B.Li,K.Tian,andZ.Lv, ‘‘Afastimageretrievalmethod
5, pp. 2350–2359, Oct. 2020. designed for network big data,’’ IEEE Trans. Ind. Inform., vol. 13, no.
[17] J. Ahmad, K. Muhammad, and S. W. Baik, ‘‘Data augmentation- 5, pp. 2350–2359, Oct. 2020.
assisted deep learning of hand-drawn partially colored sketches for
visual search,’’ PLOS ONE, vol. 12, no. 8, p. e0183838,2021.
[18] Y. M. Manu and G. K. Ravikumar Survey on Machine Learning
Based Video Analytics Techniques, Journal of Computational and
Theoretical Nanoscience, Vol. 17, 1–7, 2020

Authorized licensed use limited to: Shah and Anchor Kutchhi Engineering College. Downloaded on August 22,2024 at 08:01:29 UTC from IEEE Xplore. Restrictions apply.